From c407994dd3aeb2c5a8e5f3fa070e7436fe308fc9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ren=C3=A9=20=27Necoro=27=20Neumann?= <necoro@necoro.eu>
Date: Sat, 27 Feb 2021 23:39:51 +0100
Subject: Remove obsolete feeds from cache after 180 days

---
 CHANGELOG.md                 |  1 +
 internal/feed/cache/cache.go |  1 +
 internal/feed/cache/state.go |  9 +++++++--
 internal/feed/cache/v1.go    | 16 ++++++++++++++++
 internal/feed/cache/v2.go    |  4 ++++
 pkg/util/util.go             |  5 +++++
 6 files changed, 34 insertions(+), 2 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 7bbc284..4608d88 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -10,6 +10,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Support for JSON v1.1 feed (via [gofeed](https://github.com/mmcdole/gofeed/pull/169))
 ### Changed
 - Caches store now 1000 old entries (i.e., not included in the last fetch) at maximum. This will clean obsolete cruft and drastically reduce cache size.
+- Feeds not updated (and not part of the config) for 180 days are now automatically removed from the cache.
 
 ## [0.7.0] - 2021-02-21
 ### Changed
diff --git a/internal/feed/cache/cache.go b/internal/feed/cache/cache.go
index 171d0cb..311e871 100644
--- a/internal/feed/cache/cache.go
+++ b/internal/feed/cache/cache.go
@@ -24,6 +24,7 @@ const (
 type Impl interface {
 	cachedFeed(*feed.Feed) CachedFeed
 	transformTo(Version) (Impl, error)
+	cleanup(knownDescriptors map[feed.Descriptor]bool)
 	load(io.Reader) error
 	store(io.Writer) error
 	Version() Version
diff --git a/internal/feed/cache/state.go b/internal/feed/cache/state.go
index 2df0c74..ce6723a 100644
--- a/internal/feed/cache/state.go
+++ b/internal/feed/cache/state.go
@@ -11,6 +11,7 @@ import (
 type State struct {
 	feeds       map[string]*feed.Feed
 	cachedFeeds map[string]CachedFeed
+	knownFeeds  map[feed.Descriptor]bool
 	cache       Cache
 	cfg         *config.Config
 }
@@ -55,6 +56,7 @@ func (state *State) LoadCache(fileName string, forceNew bool) error {
 
 	for name, feed := range state.feeds {
 		state.cachedFeeds[name] = cache.cachedFeed(feed)
+		state.knownFeeds[feed.Descriptor()] = true
 	}
 
 	// state.feeds should not be used after loading the cache --> enforce a panic
@@ -64,6 +66,7 @@ func (state *State) LoadCache(fileName string, forceNew bool) error {
 }
 
 func (state *State) StoreCache(fileName string) error {
+	state.cache.cleanup(state.knownFeeds)
 	return state.cache.store(fileName)
 }
 
@@ -115,9 +118,11 @@ func (state *State) Filter() {
 }
 
 func NewState(cfg *config.Config) (*State, error) {
+	numFeeds := len(cfg.Feeds)
 	state := State{
-		feeds:       map[string]*feed.Feed{},
-		cachedFeeds: map[string]CachedFeed{},
+		feeds:       make(map[string]*feed.Feed, numFeeds),
+		cachedFeeds: make(map[string]CachedFeed, numFeeds),
+		knownFeeds:  make(map[feed.Descriptor]bool, numFeeds),
 		cache:       Cache{}, // loaded later on
 		cfg:         cfg,
 	}
diff --git a/internal/feed/cache/v1.go b/internal/feed/cache/v1.go
index 0363303..5675db5 100644
--- a/internal/feed/cache/v1.go
+++ b/internal/feed/cache/v1.go
@@ -23,6 +23,7 @@ const (
 	v1Version    Version = 1
 	startFeedId  uint64  = 1
 	maxCacheSize         = 1000
+	maxCacheDays         = 180
 )
 
 type feedId uint64
@@ -376,6 +377,21 @@ func filterItems(items []cachedItem) []cachedItem {
 	return copiedItems
 }
 
+func (cache *v1Cache) cleanup(knownDescriptors map[feed.Descriptor]bool) {
+	for descr, id := range cache.Ids {
+		if knownDescriptors[descr] {
+			// do not delete stuff still known to us
+			continue
+		}
+
+		cf := cache.Feeds[id]
+		if cf.LastCheck.IsZero() || util.Days(time.Since(cf.LastCheck)) > maxCacheDays {
+			delete(cache.Feeds, id)
+			delete(cache.Ids, descr)
+		}
+	}
+}
+
 func (cache *v1Cache) load(reader io.Reader) error {
 	decoder := gob.NewDecoder(reader)
 	return decoder.Decode(cache)
diff --git a/internal/feed/cache/v2.go b/internal/feed/cache/v2.go
index 87636b6..2abf2a8 100644
--- a/internal/feed/cache/v2.go
+++ b/internal/feed/cache/v2.go
@@ -29,6 +29,10 @@ func (cache *v2Cache) transformTo(v Version) (Impl, error) {
 	return nil, fmt.Errorf("Transformation not supported")
 }
 
+func (cache *v2Cache) cleanup(knownDescriptors map[feed.Descriptor]bool) {
+	cache.asV1().cleanup(knownDescriptors)
+}
+
 func (cache *v2Cache) Version() Version {
 	return v2Version
 }
diff --git a/pkg/util/util.go b/pkg/util/util.go
index 88d04c0..bacb494 100644
--- a/pkg/util/util.go
+++ b/pkg/util/util.go
@@ -21,3 +21,8 @@ func TimeFormat(t time.Time) string {
 
 	return t.Format(time.ANSIC)
 }
+
+// Days returns the number of days in a duration. Fraction of days are discarded.
+func Days(d time.Duration) int {
+	return int(d.Hours() / 24)
+}
-- 
cgit v1.2.3